BEHAVIORAL ASSUMPTION-BASED PREDICTION F OR HIGH-LATENCY HIDING IN MOBILE GAMES

BEHAVIORAL ASSUMPTION-BASED PREDICTION

FOR HIGH-LATENCY HIDING IN MOBILE GAMES

BEHAVIORAL ASSUMPTION-BASED PREDICTION

FOR HIGH-LATENCY HIDING IN MOBILE GAMES

Computer Graphics and CAD/CAM GroupFaculty of Electrical Engineering, Mathematics and Computer

Science

Computer Graphics and CAD/CAM GroupFaculty of Electrical Engineering, Mathematics and Computer

Science

Giliam J.P. de Carpentier Rafael Bidarra

Giliam J.P. de Carpentier Rafael Bidarra

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

The problemThe problem

• Multi-player games– Fast-paced racing

• Network connections

• Latency

• Multi-player games– Fast-paced racing

• Network connections

• Latency

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

The problemThe problem

• Multi-player games– Fast-paced racing

• Network connections

• Latency

• Multi-player games– Fast-paced racing

• Network connections

• Latency

Network

T = t1 : Transmit red position

Network

Received position red

Actual position red

T = t1 + ΔTlatency : Receive position

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

A common techniqueA common technique

• Prediction

• Prediction errors

• Dead reckoning

• Prediction

• Prediction errors

• Dead reckoning

Network

Network

Received position red at t1

Actual position red≈

Predicted position red

T = t1 + ΔTlatency : Receive position

T = t1 : Transmit red position

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

A common techniqueA common technique

• Prediction

• Prediction errors

• Dead reckoning

• Prediction

• Prediction errors

• Dead reckoning

Network

Network

Actual position red

Predicted position red

Received position red at t1

T = t1 + ΔTlatency : Receive position

T = t1 : Transmit red position

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

Mobile gamesMobile games

LAN network(PC game)

Internet(PC game)

GPRS network (Mobile game)

Network latency 1 - 5 ms 50 - 200 ms 1500 - 3500 ms

Prediction distance in a race game@ 100 km/h

0.03 – 0.14 meter

1.4 - 5.6 meter 42 – 97 meter

Prediction error, assuming a max. of 10% deviation.

< 2 cm < 0.6 meter < 10 meter

Results Dead- reckoning

Good. Medium. Interactions like collisions can be problematic.

Bad. Errors as large as the width of a track are unacceptable.

• Platform: Java, BREW, Symbian, …• Network: GPRS using HTTP/TCP/IP stack

• Platform: Java, BREW, Symbian, …• Network: GPRS using HTTP/TCP/IP stack

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

Prediction modelsPrediction models

• Standard strategy:– Only extrapolate

from older data

• Our approach:– Assume track

following behavior

– Or assume racing line optimizing behavior

• Standard strategy:– Only extrapolate

from older data

• Our approach:– Assume track

following behavior

– Or assume racing line optimizing behavior

Xt-1

Xt-2

~Xt

Xt

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

Prediction modelsPrediction models

Xt-1

Xt-2

~Xt

Xt• Standard strategy:– Only extrapolate

from older data

• Our approach:– Assume track

following behavior

– Or assume racing line optimizing behavior

• Standard strategy:– Only extrapolate

from older data

• Our approach:– Assume track

following behavior

– Or assume racing line optimizing behavior

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

Continuity Continuity

• Multiple simulations

– Running 2 or 3 real-time simulations

– Linear interpolation between simulations

– Round robin

• Multiple simulations

– Running 2 or 3 real-time simulations

– Linear interpolation between simulations

– Round robin

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

Continuity Continuity

• Multiple simulations

– Running 2 or 3 real-time simulations

– Linear interpolation between simulations

– Round robin

• Multiple simulations

– Running 2 or 3 real-time simulations

– Linear interpolation between simulations

– Round robin

simBsimA simC

time

weig

ht

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

A Test DriveA Test Drive

• Creating a testbed

• Comparing results

• Razor

• Creating a testbed

• Comparing results

• Razor

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

A Test DriveA Test Drive

• Creating a testbed

• Comparing results

• Razor

• Creating a testbed

• Comparing results

• Razor

http://www.exmachina.nl

G.J.P. de Carpentier and R. Bidarra CGAMES 2005, FranceG.J.P. de Carpentier and R. Bidarra CGAMES 2005, France

ConclusionsConclusions

• Mobile 2.5G networks: a problematic domain

• Standard dead reckoning is insufficient

• Behavioral assumptions improve prediction

• Running multiple simulations: a good fit

• Mobile 2.5G networks: a problematic domain

• Standard dead reckoning is insufficient

• Behavioral assumptions improve prediction

• Running multiple simulations: a good fit